This depends on what you want to have in the result: F-stops or T-stops. And how precise it should be. For F-stops you need to determine actual focal length and measure the diameter of the aperture as it is seen through the front element from infinite. A device called collimator might be useful, or a laser pointer mounted on a rail perpendicular to optical axis. Non ideal roundness of the aperture would be in most cases negligible, or you can measure for several times rotating your lens around its axis and calculate average.
For T-stops you can use an incident light meter, or better a 'booster' for on-ground-glass metering (you have to calibrate it with your ground-glass first).
For practical application it should be enough to:
1) believe that maximum aperture stated by manufacturer is true
2) assume F-stops are equal T-stops
3) meter relative, so calibration with the ground glass is not needed, 1 EV (or whatever else) T-stops from maximum aperture